A system and method of monitoring and diagnosing on-line multivariate process variable. Multivariate process data includes multiple process variables each having a multiple observations. On-line process data is collected from a process control system when the process is on-line. The on-line process data includes multiple observations of multiple process variables. A multivariate statistical analysis represents the operation of the process based on a set of the on-line process data. The representation a result. The representation and the set of on-line process data are stored. An output is generated based on a parameter of the representation. The parameter includes a result generated by the representation of the operation of the process, a process variable used to generate the representation of the operation of the process and/or the set of collected on-line process data.
Legal claims defining the scope of protection. Each claim is shown in both the original legal language and a plain English translation.
1. A system for monitoring and diagnosing on-line multivariate process variable data in a process plant, wherein the multivariate process data comprises a plurality of process variables each having a plurality of observations, the system comprising: a data collection tool adapted to collect on-line process data from a process control system within the process plant when the process is on-line, wherein the collected on-line process data comprises a plurality of observations of a plurality of process variables, and wherein the plurality of observations of a set of collected on-line process data comprises a first data space having a plurality of dimensions; an analysis tool comprising a multivariate statistical analysis engine adapted to represent the operation of the process within a second data space having fewer dimensions than the first data space based on a set of collected on-line process data comprising a measure of the operation of the process when the process is on-line, wherein the representation of the operation of the process is adapted to be executed to generate a result, and wherein the analysis tool is adapted to store the representation of the operation of the process and the set of collected on-line process data; and a monitoring tool adapted to generate an output based on a parameter of the representation of the operation of the process, wherein the parameter of the representation of the operation of the process comprises one or more of a result generated by the representation of the operation of the process, a process variable used to generate the representation of the operation of the process and the set of collected on-line process data.
A system monitors and diagnoses process variable data in a process plant. It includes a data collection tool to gather real-time process data (multiple variables, multiple observations) from a process control system. This data exists in a high-dimensional space. An analysis tool uses multivariate statistical analysis to represent the process's operation in a lower-dimensional space. This representation, when executed, produces a result. The system stores this representation and the collected data. A monitoring tool generates an output based on parameters of the representation, such as the result, a process variable used, or the collected data itself.
2. The system of claim 1 , wherein the monitoring tool comprises an execution tool adapted to execute the representation of the operation of the process to generate a score.
The system described in claim 1 also features a monitoring tool that includes an execution tool. This execution tool runs the process representation to produce a "score." This score reflects the operational state of the process, based on the analysis.
3. The system of claim 2 , wherein the score comprises a representation of one or more observations in the second data space.
Building on the system described in claim 2, the "score" generated by the execution tool is a representation of one or more observations from the process data within the reduced-dimension data space created by the multivariate statistical analysis. Essentially, it's a simplified view of the current process state.
4. The system of claim 3 , wherein the monitoring tool is adapted to project the observations from the set of collected data onto the second data space to generate one or more scores.
Further detailing the system described in claim 3, the monitoring tool projects the original, high-dimensional process data onto the lower-dimensional space to calculate the scores. This projection allows the system to assess the process's current state in terms of the reduced-dimension representation.
5. The system of claim 3 , wherein the multivariate statistical analysis engine is adapted to represent the operation of the process as a plurality of principal components defining the second data space, wherein each principal component corresponds to an amount of variation in the set of collected process data.
In the system outlined in claim 3, the multivariate statistical analysis represents the process as a set of principal components. These components define the lower-dimensional space, with each component corresponding to a specific amount of variation within the original process data. The system uses these principal components to understand the dominant patterns in the process data.
6. The system of claim 2 , wherein the monitoring tool is adapted to diagnose the process using the statistical output.
Expanding upon the system described in claim 2, the monitoring tool uses the statistical output (the score) to diagnose problems within the process. By analyzing the score, the system can detect and identify potential issues in the process.
7. The system of claim 2 , wherein the monitoring tool is adapted to determine the presence of an abnormal process condition if the score exceeds a threshold value.
Building on the system described in claim 2, the monitoring tool determines if an abnormal condition exists. It does this by comparing the calculated "score" to a predefined threshold. If the score exceeds this threshold, the system flags the process as being in an abnormal state.
8. The system of claim 7 , wherein the analysis tool is adapted to determine the threshold value based upon one or more process variable limits of the plurality of process variables.
Within the system described in claim 7, the analysis tool determines the threshold value used to detect abnormal conditions based on process variable limits. These limits define acceptable ranges for the process variables.
9. The system of claim 8 , wherein the one or more process variable limits of the plurality of process variables comprises one or more of the group comprising: a statistical process variable limit and a physical process variable limit.
Clarifying the system described in claim 8, the process variable limits used to determine the threshold can be either statistical limits (derived from process data) or physical limits (based on equipment constraints). The system can use either type of limit, or a combination of both, to set the threshold.
10. The system of claim 7 , wherein the monitoring tool is adapted to generate an alarm based on the presence of an abnormal process condition.
Further describing the system in claim 7, when the monitoring tool detects an abnormal process condition (score exceeds the threshold), it generates an alarm. This alarm signals operators that intervention may be required.
11. The system of claim 1 , wherein the monitoring tool is adapted to generate a normalized process variable based on a process variable used to generate the representation of the operation of the process.
In the monitoring system described in claim 1, the monitoring tool generates a "normalized" process variable based on a process variable used to create the process representation. This normalization helps to standardize the different process variables for easier comparison and analysis.
12. The system of claim 11 , wherein each process variable used to generate the representation of the operation of the process comprises a known process variable limit and a known target value, wherein the monitoring tool comprises a normalization tool adapted to normalize each process variable based on the known process variable limit and the known target.
Expanding on claim 11, in the system where process variables used in the model have known limits and target values, the monitoring tool includes a normalization tool. This tool normalizes each variable based on its known limit and target, allowing for comparison of variables with different scales and units.
13. The system of claim 12 , wherein the known process variable limit comprises a known upper process variable limit and a known lower process variable limit.
Further explaining claim 12, the known process variable limit used for normalization consists of a known upper limit and a known lower limit. The normalization uses both these limits, along with the target value, to rescale the process variable.
14. The system of claim 11 , wherein at least one of the process variables used to generate the representation of the operation of the process comprises an unknown process variable limit, wherein the monitoring tool comprises a normalization tool adapted to calculate the unknown process variable limit based on the maximum deviation of an observation from a mean of a set of collected process variable data, wherein the observation is within a set of collected process variable data comprising a measure of the at least one process variable when the process is on-line and operating normally, and normalize the at least one process variable based on the calculated unknown process variable limit and based on the mean of the set of collected process variable data to generate a process variable limit for the at least one process variable.
Building on the system of claim 11, if a process variable lacks known limits, the system calculates an "unknown" process variable limit. The normalization tool finds the maximum deviation from the mean during normal operation. It then normalizes the variable using this calculated limit and the mean. This allows the system to handle variables with undefined boundaries.
15. The system of claim 11 , wherein at least one of the process variables used to generate the representation of the operation of the process comprises an unknown process variable limit, wherein the monitoring tool comprises a normalization tool adapted to normalize the at least one process variable based on a mean of a set of collected process variable data comprising a measure of the at least one process variable when the process is on-line and operating normally and based on a variance of the set of collected process variable data to generate a process variable limit for the at least one process variable.
Continuing from claim 11, where a process variable has unknown limits, the system normalizes using the mean and variance of the variable during normal operation. The normalization tool estimates a process variable limit from the data's mean and variance. This statistical approach allows normalization even with missing limit information.
16. The system of claim 15 , wherein the normalization tool is adapted to determine a statistical control limit based on the variance of the set of collected process variable data and adapted to normalize the at least one process variable based on the statistical control limit.
Regarding the system of claim 15, the normalization tool determines a statistical control limit based on the variance of the collected data. This limit is then used to normalize the process variable. This creates a statistically relevant normalization.
17. The system of claim 11 , wherein the monitoring tool is adapted to generate a common process variable limit for the plurality of process variables based on the normalized process variable, and adapted to monitor each process variable in relation to the common process variable limit.
Extending the system of claim 11, the monitoring tool calculates a "common" process variable limit based on the normalized variables. Then, it monitors each variable against this common limit. This enables unified monitoring across different process variables.
18. The system of claim 11 , wherein the monitoring tool is adapted to determine the presence of an abnormal process condition if the process variable exceeds the common process variable limit.
Building on the system of claim 11, if a process variable exceeds the common limit, the monitoring tool detects an abnormal condition. This allows for a standardized comparison of all process variables.
19. The system of claim 18 , wherein the monitoring tool is adapted to generate an alarm based on the presence of an abnormal process condition.
The system of claim 18 takes action by generating an alarm when an abnormal condition is detected, signaling that a process variable is outside the acceptable range defined by the common limit.
20. The system of claim 18 , wherein the monitoring tool is adapted to monitor the process variable in relation to the common process variable limit based upon a result generated by the representation of the operation of the process.
Expanding on claim 18, the monitoring tool uses the result generated by the process model to monitor process variables in relation to the common limit. This means that the overall process state, as captured by the model, influences how the system evaluates individual variable deviations.
21. The system of claim 1 , wherein the monitoring tool is adapted to generate a process variable index based on the set of collected on-line process data and based on a component of the second data space, wherein the process variable index indicates an amount of deviation of one or more of the plurality of process variables and adapted to verify the representation of the operation of the process based on the process variable index.
In the system described in claim 1, the monitoring tool generates a "process variable index" based on collected process data and a component from the reduced-dimension data space. This index indicates how much a process variable deviates from its normal behavior and it's used to verify the process representation's accuracy.
22. The system of claim 21 , wherein the monitoring tool comprises an index tool adapted to determine the process variable index based on a trace of the set of collected process data in the first data space and a component of the second data space, wherein the component of the second data space comprises the component used to determine the representation of the operation of the process.
Building on claim 21, the system includes an index tool that calculates the process variable index using the "trace" of the process data in its original high-dimensional space and a component of the reduced-dimension space. The component used is the one that was used to create the model's process representation.
23. The system of claim 22 , wherein the analysis tool is adapted to generate a plurality of principal components within the first data space that define the second data space, wherein each principal component corresponds to an amount of variation in the set of collected process data and wherein the component of the second data space comprises a principal component used to define the second data space.
In the system described in claim 22, the analysis tool creates a set of "principal components" in the original, high-dimensional data space. These components define the reduced-dimension space. Each component represents a different amount of variation in the process data. The "component" used in the index calculation is one of these principal components.
24. The system of claim 21 , wherein the multivariate statistical analysis engine is adapted to represent the operation of the process based on a first set of collected on-line process data comprising a first measure of the operation of the process when the process is on-line, and adapted to represent the operation of the process based on a second set of collected on-line process data comprising a second measure of the operation of the process when the process is on-line, wherein the monitoring tool is adapted to generate a first process variable index based on the first set of collected on-line process data and based on a component used to determine the first representation of the operation of the process and adapted to generate a second process variable index based on the second set of collected on-line process data and based on a component used to determine the first representation of the operation of the process, and wherein the monitoring tool comprises a verification tool adapted to compare the first and second process variables to determine whether the first and second process variables are on the same order if the first and second representations of the operation of the process represent the same operation of the process.
In the monitoring system of claim 21, the multivariate statistical analysis creates two representations of the process based on two sets of data. The monitoring tool computes two process variable indexes, one for each set of data, using the components that defined the first process representation. It then compares the indexes to see if they're of the same magnitude, thus checking that the process representations are similar.
25. The system of claim 24 , wherein the analysis tool comprises a first multivariate statistical analysis engine adapted to represent the operation of the process based on the first set of collected on-line process data comprising the first measure of the operation of the process when the process is on-line, and a second multivariate statistical analysis engine adapted to represent the operation of the process based on the set of collected on-line process data comprising the second measure of the operation of the process when the process is on-line.
Building on claim 24, the system uses two separate multivariate statistical analysis engines: one for each set of process data used to create the two process representations. This enables the system to independently model the process from two different data sources or time periods.
26. The system of claim 21 , wherein the monitoring tool is adapted to store the process variable index with the representation of the operation of the process and the set of collected on-line process data.
Expanding on claim 21, the system stores the process variable index alongside the process representation and the collected data used to create it. This storage allows for later analysis, comparison, and verification of the model's performance.
27. The system of claim 21 , wherein the index tool is adapted to determine the process variable index as a calculation of: PV j rating = 1 trS ∑ i = 1 k σ i 2 p i , j 2 wherein: PV jrating =the process variable index of a jth process variable, S=an n×n auto-scaled covariance matrix of the set of collected on-line process data, n=plurality of dimensions of the first data space, trS=a trace of S, k=the plurality of dimensions of the second data space, σ i =a principal eigenvalue of S, p i =a principal eigenvector of S, and p i,j =the jth component of p i .
In the system defined by claim 21, the process variable index is calculated using a specific formula. The formula involves the trace of the auto-scaled covariance matrix of the data, the eigenvalues of the covariance matrix, and the components of the principal eigenvectors.
28. The system of claim 1 , further comprising a display tool adapted to generate a visualization of the output generated by the monitoring tool.
Adding to the monitoring system of claim 1, a display tool generates a visualization of the output produced by the monitoring tool. This helps operators understand the process state and any potential issues.
29. The system of claim 28 , wherein the display tool is adapted to generate a visualization of a score generated by the representation of the operation of the process in relation to a component used to define the second data space.
Expanding on claim 28, the display tool generates a visualization of the score produced by the process representation, shown in relation to the components that define the reduced-dimensional space. This allows users to see how the process is behaving in the context of the underlying statistical model.
30. The system of claim 29 , wherein the visualization of a score generated by the representation of the operation of the process in relation to a component used to define the second data space comprises a visualization of an on-line score generated by a representation of an on-line operation of the process when the process is operating on-line in relation to a component used to define the second data space.
Elaborating on claim 29, the score visualization includes the online score generated by the process representation while the process is running in real-time, shown in relation to the components of the reduced data space. This gives operators real-time feedback on process performance relative to the model.
31. The system of claim 29 , wherein the visualization of a score generated by the representation of the operation of the process in relation to a component used to define the second data space further comprises a visualization of one or more of the plurality of process variables in relation to the component used to define the second data space.
Building on claim 29, the visualization of the score also includes one or more of the process variables shown in relation to the components used to define the reduced space. This allows operators to see which process variables are contributing most to the score and driving the overall process behavior.
32. The system of claim 28 , wherein the visualization comprises a plot of a score generated by the representation of the operation of the process versus time.
Expanding on claim 28, the visualization includes a plot of the score over time. This allows users to track the process's performance trends and identify potential problems before they escalate.
33. The system of claim 28 , wherein the visualization comprises a plot of a first score generated by the representation of the operation of the process versus a second score generated by the representation of the operation of the process.
Expanding on claim 28, the visualization includes a plot of one score versus another score generated by the process model. This can help reveal relationships between different aspects of the process.
34. The system of claim 28 , wherein the display tool is adapted to generate a visualization of one or more of the plurality of process variables and a process variable index associated with each of the one or more plurality of process variables indicating an amount of deviation of the associated process variable.
Expanding on claim 28, the display tool visualizes one or more of the process variables alongside the process variable index. This index indicates the amount of deviation of the variable, giving operators a clear view of which variables are most problematic.
35. The system of claim 28 , wherein the display tool is adapted to generate a visualization of one or more of the plurality of process variables in relation to a process variable limit associated with each process variable.
Expanding on claim 28, the display tool visualizes one or more process variables in relation to their associated process variable limits. This allows operators to easily see if a variable is within its acceptable range.
36. The system of claim 1 , wherein the monitoring tool is adapted to generate one or more outputs selected from the group consisting of: a statistical output, a normalized process variable, and a process variable rating, wherein the statistical output comprises representations of observations of on-line process data in the second data space based on on-line process data having the largest data variation, wherein the normalize process variable comprises a process variable normalized as a function of a data set collected for the process variable and used to train the representation of operation of the process when the process is operating normally, and wherein the process variable rating comprises a rating of a process variable according to the variability of the process variable.
In the system of claim 1, the monitoring tool generates one or more of the following outputs: a statistical output, a normalized process variable, and a process variable rating. The statistical output shows the representation of the data with the largest variations; the normalized variable is a standardized version based on the training data; and the variable rating indicates each variable's variability.
37. A method of monitoring and diagnosing on-line multivariate process variable data in a process plant, wherein the multivariate process data comprises a plurality of process variables each having a plurality of observations, the method comprising: collecting on-line process data from a process control system within the process plant when the process is on-line, wherein the collected on-line process data comprises a plurality of observations of a plurality of process variables, and wherein the plurality of observations of a set of collected on-line process data comprises a first data space having a plurality of dimensions; performing a multivariate statistical analysis, by a multivariate statistical analysis engine, to represent the operation of the process within a second data space having fewer dimensions than the first data space based on a set of collected on-line process data comprising a measure of the operation of the process when the process is on-line, wherein the representation of the operation of the process is adapted to be executed to generate a result; storing the representation of the operation of the process and the set of collected on-line process data; and generating an output based on a parameter of the representation of the operation of the process, wherein the parameter of the representation of the operation of the process comprises one or more of a result generated by the representation of the operation of the process, a process variable used to generate the representation of the operation of the process and the set of collected on-line process data.
A method monitors and diagnoses process variable data in a process plant. First, it collects real-time process data (multiple variables, multiple observations) from a process control system, existing in a high-dimensional space. Next, it uses multivariate statistical analysis to represent the process's operation in a lower-dimensional space, generating a result upon execution. This representation and the collected data are stored. Finally, an output is generated based on parameters of the representation, such as the result, a process variable used, or the collected data itself.
38. The method of claim 37 , wherein generating an output based on a parameter of the representation of the operation of the process comprises executing the representation of the operation of the process to generate a score.
The method described in claim 37 generates an output by executing the process representation to produce a "score." This score reflects the operational state of the process, based on the analysis.
39. The method of claim 38 , wherein the score comprises a representation of one or more observations in the second data space.
Building on the method described in claim 38, the "score" generated is a representation of observations from the process data within the reduced-dimension data space. Essentially, it's a simplified view of the current process state.
40. The method of claim 39 , further comprising projecting the observations from the set of collected data onto the second data space to generate one or more scores.
The method described in claim 39 projects the original, high-dimensional process data onto the lower-dimensional space to calculate the scores. This projection allows assessment of the process's current state in the reduced-dimension representation.
41. The method of claim 39 , wherein performing a multivariate statistical analysis to represent the operation of the process comprises generating a plurality of principal components within the first data space, wherein each principal component corresponds to an amount of variation in the set of collected process data, and wherein the principal components define the second data space.
This invention relates to process monitoring and control systems, specifically methods for analyzing process data to detect anomalies or deviations. The problem addressed is the need for efficient and accurate representation of complex process data to identify operational issues in real-time. The method involves collecting process data from sensors or other monitoring devices, where the data represents various parameters of an industrial or manufacturing process. A multivariate statistical analysis is performed to transform the collected data into a reduced-dimensional representation. This analysis generates a plurality of principal components within an initial data space, where each principal component quantifies a distinct amount of variation in the process data. The principal components collectively define a new, lower-dimensional data space that captures the most significant patterns in the original data. This transformation simplifies the data while preserving critical information, enabling more effective anomaly detection and process monitoring. The method may also include comparing the transformed data to historical or expected values to identify deviations, triggering alerts or adjustments as needed. The approach improves process efficiency and reduces downtime by enabling early detection of potential issues.
42. The method of claim 38 , further comprising diagnosing the process using the statistical output.
Expanding upon the method described in claim 38, the method diagnoses problems within the process using the statistical output (the score), enabling detection and identification of potential process issues.
43. The method of claim 38 , further comprising determining the presence of an abnormal process condition if the score exceeds a threshold value.
Building on the method described in claim 38, the method determines if an abnormal condition exists by comparing the "score" to a threshold. If the score exceeds it, the process is flagged as abnormal.
44. The method of claim 43 , further comprising determining the threshold value based upon one or more process variable limits of the plurality of process variables.
Within the method described in claim 43, the threshold is determined based on process variable limits, which define acceptable ranges for the process variables.
45. The method of claim 44 , wherein the one or more process variable limits of the plurality of process variables comprises one or more of the group comprising: a statistical process variable limit and a physical process variable limit.
Clarifying the method described in claim 44, process variable limits used to determine the threshold can be statistical (derived from process data) or physical (equipment constraints), allowing use of either type or both.
46. The method of claim 13 , further comprising generating an alarm based on the presence of an abnormal process condition.
Describing the method from claim 13, an alarm is generated when an abnormal process condition is detected (score exceeds the threshold), signaling potential intervention. *NOTE: Claim 46 references claim 13, but this appears to be a typo and should reference Claim 43 instead.*
47. The method of claim 37 , wherein generating an output based on a parameter of the representation of the operation of the process comprises generating a normalized process variable based on a process variable used to generate the representation of the operation of the process.
In the monitoring method of claim 37, generating an output involves creating a "normalized" process variable based on a variable used to create the process representation. This standardization aids comparison and analysis.
48. The method of claim 47 , wherein each process variable used to generate the representation of the operation of the process comprises a known process variable limit and a known target value, wherein generating a normalized process variable comprises normalizing each process variable based on the known process variable limit and the known target.
Expanding on claim 47, if process variables have known limits and target values, the method normalizes each variable based on its known limit and target.
49. The method of claim 48 , wherein the known process variable limit comprises a known upper process variable limit and a known lower process variable limit.
Further explaining claim 48, the known process variable limit used for normalization consists of a known upper and lower limit.
50. The method of claim 47 , wherein at least one of the process variables used to generate the representation of the operation of the process comprises an unknown process variable limit, wherein generating a normalized process variable comprises: calculating the unknown process variable limit based on the maximum deviation of an observation from a mean of a set of collected process variable data, wherein the observation is within a set of collected process variable data comprising a measure of the at least one process variable when the process is on-line and operating normally; and normalizing the at least one process variable based on the calculated unknown process variable limit and based on the mean of the set of collected process variable data to generate a process variable limit for the at least one process variable.
Building on claim 47, if a process variable lacks known limits, the method calculates an "unknown" limit, finding the maximum deviation from the mean during normal operation, and normalizes using this limit and the mean.
51. The method of claim 47 , wherein at least one of the process variables used to generate the representation of the operation of the process comprises an unknown process variable limit, wherein generating a normalized process variable comprises: normalizing the at least one process variable based on a mean of a set of collected process variable data comprising a measure of the at least one process variable when the process is on-line and operating normally and based on a variance of the set of collected process variable data to generate a process variable limit for the at least one process variable.
Continuing from claim 47, where a process variable has unknown limits, the method normalizes using the mean and variance of the variable during normal operation to estimate a process variable limit.
52. The method of claim 51 , wherein generating a normalized process variable comprises: determining a statistical control limit based on the variance of the set of collected process variable data; and normalizing the at least one process variable based on the statistical control limit.
Regarding the method of claim 51, a statistical control limit is determined based on the variance, then used to normalize the variable, creating a statistically relevant normalization.
53. The method of claim 47 , further comprising: generating a common process variable limit for the plurality of process variables based on the normalized process variable; and monitoring each process variable in relation to the common process variable limit.
Extending the method of claim 47, a "common" process variable limit is calculated, and each variable is monitored against this limit for unified monitoring.
54. The method of claim 47 , further comprising determining the presence of an abnormal process condition if the process variable exceeds the common process variable limit.
Building on the method of claim 47, an abnormal condition is detected if a variable exceeds the common limit.
55. The method of claim 54 , further comprising generating an alarm based on the presence of an abnormal process condition.
The method of claim 54 generates an alarm when an abnormal condition is detected, indicating a variable outside the acceptable range defined by the common limit.
56. The method of claim 54 , further comprising monitoring the process variable in relation to the common process variable limit based upon a result generated by the representation of the operation of the process.
Expanding on claim 54, the process model's result is used to monitor variables in relation to the common limit, meaning the overall process state influences variable deviation evaluation.
57. The method of claim 37 , wherein generating an output based on a parameter of the representation of the operation of the process comprises generating a process variable index based on the set of collected on-line process data and based on a component of the second data space, wherein the process variable index indicates an amount of deviation of one or more of the plurality of process variables.
In the method of claim 37, output generation involves a "process variable index" based on collected data and a component from the reduced space, indicating variable deviation.
58. The method of claim 57 , further comprising determining the process variable index based on a trace of the set of collected process data in the first data space and a component of the second data space, wherein the component of the second data space comprises the component used to determine the representation of the operation of the process.
Building on claim 57, the index is calculated using the "trace" of the process data in its original space and a component of the reduced space, used to create the model's representation.
59. The method of claim 58 , wherein a plurality of principal components within the first data space define the second data space, wherein each principal component corresponds to an amount of variation in the set of collected process data and wherein the component of the second data space comprises a principal component used to define the second data space.
In the method described in claim 58, "principal components" define the reduced space, each representing data variation, with the "component" being one of these.
60. The method of claim 57 , wherein performing a multivariate statistical analysis comprises: representing the operation of the process based on a first set of collected on-line process data comprising a first measure of the operation of the process when the process is on-line; and representing the operation of the process based on a second set of collected on-line process data comprising a second measure of the operation of the process when the process is on-line, and wherein generating a process variable index comprises: generating a first process variable index based on the first set of collected on-line process data and based on a component used to determine the first representation of the operation of the process; and generating a second process variable index based on the second set of collected on-line process data and based on a component used to determine the first representation of the operation of the process, the method further comprising comparing the first and second process variables to determine whether the first and second process variables are on the same order if the first and second representations of the operation of the process represent the same operation of the process.
In the method of claim 57, the statistical analysis creates two representations from two data sets. Two indexes are computed, one for each data set, using components that defined the first representation. The indexes are compared to see if they are similar, checking process representation similarity.
61. The method of claim 60 , wherein representing the operation of the process based on a first set of collected on-line process data comprises performing a first multivariate statistical analysis to represent the operation of the process based on the first set of collected on-line process data comprising the first measure of the operation of the process when the process is on-line, and wherein representing the operation of the process based on a second set of collected on-line process data comprises performing a second multivariate statistical analysis to represent the operation of the process based on the set of collected on-line process data comprising the second measure of the operation of the process when the process is on-line.
Building on claim 60, the method employs two statistical analysis steps: one for each set of process data to model the process from different sources.
62. The method of claim 57 , further comprising storing the process variable index with the representation of the operation of the process and the set of collected on-line process data.
Expanding on claim 57, the method stores the index with the representation and data used to create it for later analysis, comparison, and verification.
63. The method of claim 1 , further comprising generating a visualization of the output generated by the monitoring tool.
The method in claim 1 includes a step to visualize the output generated by the monitoring.
64. The method of claim 63 , wherein generating a visualization comprises generating a visualization of a score generated by the representation of the operation of the process in relation to a component used to define the second data space.
For the method of claim 63, visualization includes the score in relation to the components that define the data space.
65. The method of claim 64 , wherein the visualization of a score generated by the representation of the operation of the process in relation to a component used to define the second data space comprises a visualization of an on-line score generated by a representation of an on-line operation of the process when the process is operating on-line in relation to a component used to define the second data space.
Method of claim 64, visualization includes the online score generated when the process is running.
66. The method of claim 64 , wherein the visualization of a score generated by the representation of the operation of the process in relation to a component used to define the second data space further comprises a visualization of one or more of the plurality of process variables in relation to the component used to define the second data space.
Expanding claim 64, the score visualization includes process variables in relation to the components used to define the space.
67. The method of claim 63 , wherein the visualization comprises a plot of a score generated by the representation of the operation of the process versus time.
Expanding claim 63, the visualization includes a plot of the score versus time.
68. The method of claim 63 , wherein the visualization comprises a plot of a first score generated by the representation of the operation of the process versus a second score generated by the representation of the operation of the process.
Expanding claim 63, visualization includes a plot of one score vs another.
69. The method of claim 63 , wherein generating a visualization comprises generating a visualization of one or more of the plurality of process variables and a process variable index associated with each of the one or more plurality of process variables indicating an amount of deviation of the associated process variable.
Expanding claim 63, the method includes visualizing variables alongside their indexes which indicate their deviations.
70. The method of claim 63 , wherein generating a visualization comprises generating a visualization of one or more of the plurality of process variables in relation to a process variable limit associated with each process variable.
Expanding claim 63, the visualization includes process variables in relation to their limits.
71. A system for monitoring and diagnosing on-line multivariate process variable data in a process plant, wherein the multivariate process data comprises a plurality of process variables each having a plurality of observations, the system comprising: a data collection tool adapted to collect on-line process data from a process control system within the process plant when the process is on-line, wherein the collected on-line process data comprises a plurality of observations of a plurality of process variables, and wherein the plurality of observations of a set of collected on-line process data comprises a first data space having a plurality of dimensions; an analysis tool comprising a multivariate statistical analysis engine adapted to represent the operation of the process within a second data space having fewer dimensions than the first data space based on a set of collected on-line process data comprising a measure of the operation of the process when the process is on-line, wherein the representation of the operation of the process is adapted to be executed to generate a result, and wherein the analysis tool is adapted to store the representation of the operation of the process and the set of collected on-line process data; a normalization tool adapted to normalize two or more of the plurality of process variables; and a monitoring tool adapted to determine a process variable limit based on the normalized process variables and to monitor one or more of the process variables in relation to the process variable limit.
A system to monitor and diagnose process data normalizes two or more process variables. The system consists of: a data collection component to gather real-time process data, an analysis tool that creates a model of the process using statistical analysis, a normalization tool that normalizes two or more process variables and a monitoring tool that determines a variable limit based on the normalized variables and monitors process variables against this limit.
72. The system of claim 71 , wherein each process variable used to generate the representation of the operation of the process comprises a known process variable limit and a known target value, and wherein the normalization tool is adapted to normalize two or more of the plurality of process variables as a calculation of: PV * = 100 % × n × ( PV - T ) 2 CL 2 wherein: PV*=a normalized process variable, T=the known target value, CL=|UCL−T|=|LCL−T|, UCL=a known upper process variable limit, LCL=a known lower process variable limit, and 0<n<1.
The system described in claim 71 includes a normalization tool that normalizes process variables when each variable has a known limit and target value. The normalization involves calculating a normalized variable (PV*) using the formula PV* = 100% * n * (PV - T)^2 / CL^2, where T is the target value, CL is the distance between the target and the upper or lower control limit, and n is a factor between 0 and 1.
73. The system of claim 71 , wherein at least one of the plurality of process variables used to generate the representation of the operation of the process comprises an unknown process variable limit, wherein the normalization tool is adapted to normalize two or more of the plurality of process variables as a calculation of: PV * = 100 % × n × ( PV - M PV ) 2 CL 2 wherein: PV*=a normalized process variable, CL==|PV−M PV | max in τ{PV}, PV=an observation of a process variable, τ{PV}=a set of collected on-line process data comprising the plurality of observations of the process variable when the process is operating normally, M PV =the mean of τ{PV}, and 0<n<1.
The system described in claim 71 includes a normalization tool that normalizes process variables when at least one variable has an unknown limit. It uses the formula PV* = 100% * n * (PV - MPV)^2 / CL^2, where MPV is the mean of the variable when the process is operating normally, CL is the maximum deviation from the mean during normal operation, and n is a factor between 0 and 1.
74. The system of claim 71 , wherein at least one of the plurality of process variables used to generate the representation of the operation of the process comprises an unknown process variable limit, wherein the normalization tool is adapted to normalize two or more of the plurality of process variables as a calculation of: PV * = 100 % × n × ( PV - M PV ) 2 k 2 σ 2 wherein: PV*=a normalized process variable, kσ=a statistical process control limit, k=1, 2, . . . , n, PV=an observation of a process variable, τ{PV}=a set of collected on-line process data comprising the plurality of observations of the process variable when the process is operating normally, M PV =the mean of τ{PV}, σ=a variance of τ{PV}, and 0<n<1.
The system described in claim 71 includes a normalization tool that normalizes process variables when at least one has an unknown limit. It uses the formula PV* = 100% * n * (PV - MPV)^2 / (k^2 * sigma^2), where MPV is the mean of the variable, sigma is the variance, k is a factor (1, 2, etc), and n is a factor between 0 and 1. The value k*sigma represents a statistical process control limit.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
March 20, 2007
July 16, 2013
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.